Google Earth Engine Setup Guide

Overview

This directory contains the Google Earth Engine JavaScript code for satellite-based water balance computation and AWD suitability assessment. The script processes CHIRPS rainfall, MODIS evapotranspiration, and SoilGrids soil data to produce dekad-level (10-day) water balance maps.

Prerequisites

Google Earth Engine Account: Sign up at https://earthengine.google.com/
Google Drive Account: For storing exported results
Study Area Assets: Define rice extent boundaries (see "Preparing Study Area Data" below)

Quick Start

1. Open Script in GEE Editor

Go to Google Earth Engine Code Editor
Create a new script and copy the contents of water_balance_suitability.js
Or import the shared script (coming soon)

2. Configure Study Area

At the top of the script, modify the configuration block:

// CONFIGURATION
var CONFIG = {
  year: 2022,  // Analysis year
  season: {dekad_start: 10, dekad_end: 28},  // Season (dekad 10-28 = May-Sep)
  exclusion_windows: {exclude_first: 2, exclude_last: 1},  // Exclude establishment/harvest phases
  deficit_thresholds: [-25, -50, -70, -90, -110, -130, -150],  // Deficit thresholds (mm)
  irrigation_threshold: 5.0,  // Min rainfall threshold (mm)
  output_scale: 500,  // Output resolution (meters)
  max_pixels: 1e13  // Max processing area
};

// STUDY AREA (replace with your rice extent)
var rice_map = ee.Image("users/your-username/japan_rice_2020");

3. Load Study Area Data

Option A: Use User Asset (Recommended)

If you have a pre-processed rice extent map:

var rice_map = ee.Image("users/[your-username]/japan_rice_2020");

Where the image contains values 1=rice, 0=other.

Option B: Use Existing MODIS Crop Classification

// MODIS Global Crop Classification (MOD12Q1)
var lulc = ee.ImageCollection("MODIS/061/MCD12Q1")
  .filterDate(CONFIG.year + "-01-01", CONFIG.year + "-12-31")
  .first()
  .select("LC_Type1");

// Class 10 = Croplands (includes rice)
var rice_map = lulc.eq(10);

Option C: Use ESA WorldCover

var worldcover = ee.ImageCollection("ESA/WorldCover/v100")
  .filterDate(CONFIG.year + "-01-01", CONFIG.year + "-12-31")
  .first();

// Class 40 = Croplands
var rice_map = worldcover.eq(40);

4. Run the Script

Click Run in the top-right corner
Monitor progress in the Tasks panel
Once data is ready, the script will start exporting results (see Tasks tab)

5. Download Results

Click Tasks tab in the GEE Editor (upper right)
For each export task:
- Click Run to start export
- Check Google Drive folder specified in config (default: "AWD")
- Download .tif files

Expected output files:

Japan_AWD_Suitability_All_Thresholds.tif (multi-band image)

Output Specification

File Format

Format: GeoTIFF (Cloud-optimized)
Bands: 7 bands (one per threshold)
- Band 1: suitability_-25 (most lenient threshold)
- Band 2: suitability_-50
- ...
- Band 7: suitability_-150 (most stringent threshold)
Values: 1-3 (suitability class)
- 1 = Low suitability (<33% suitable dekads)
- 2 = Moderate suitability (33-66% suitable dekads)
- 3 = High suitability (>66% suitable dekads)
Projection: EPSG:4326 (WGS84)
Resolution: 500 m (as configured)

Interpretation

For each pixel, values across bands show how suitability classification changes with deficit threshold stringency:

Threshold:       -25mm  -50mm  -70mm  -90mm  -110mm -130mm -150mm
Example Pixel:    3      3      2      2      1      1      1

This pixel has high suitability under lenient thresholds but drops to low suitability as threshold becomes more stringent. Use as basis for robustness assessment.

Data Sources

1. Rainfall: CHIRPS

Collection: UCSB-CHG/CHIRPS/DAILY
Resolution: 5 km
Temporal Coverage: 1981-present (daily)
Processing in Script:
- Filter to study area and season dates
- Sum 10-day windows to dekad totals
- Output units: mm/dekad

// Example access:
var chirps = ee.ImageCollection("UCSB-CHG/CHIRPS/DAILY")
  .filterDate("2022-05-01", "2022-09-30")
  .filterBounds(study_area_boundary);

2. Evapotranspiration: MODIS MOD16A2

Collection: MODIS/061/MOD16A2
Resolution: 500 m
Temporal Coverage: 2000-present (8-day)
Processing in Script:
- Reproject to 500m study grid
- Convert from 0.1 mm/8-day scale to mm/dekad
- Weighted overlap interpolation when dekad boundary crosses 8-day tile boundary
- Output units: mm/dekad

// Example access:
var modis_et = ee.ImageCollection("MODIS/061/MOD16A2")
  .filterDate("2022-05-01", "2022-09-30");

3. Soil Properties: SoilGrids

Collections:
- projects/soilgrids-isric/clay_mean (clay percentage 0-5cm)
- projects/soilgrids-isric/sand_mean (sand percentage 0-5cm)
Resolution: 250 m
Processing in Script:
- Calculate clay + sand percentages
- Classify into 4 texture classes
- Map texture class to percolation rate (3-12 mm/day)
- Convert to dekad totals
- Output units: mm/dekad

// Example access:
var clay = ee.Image("projects/soilgrids-isric/clay_mean").divide(10);  // 0-100%

4. Elevation: SRTM

Collection: USGS/SRTMGL1_Ellip
Resolution: 30 m
Processing in Script:
- Used to compute slope for biophysical constraints
- Slope threshold: <10° for feasibility
- Output units: degrees

// Example access:
var dem = ee.Image("USGS/SRTMGL1_Ellip");
var slope = ee.Terrain.slope(dem);

5. Rice Extent Map

User-provided or derived from MODIS/ESA classification
Expected values: 1 (rice) or 0 (other)
Resolution: Any (will be resampled to 500m)

Advanced Configuration

Changing Deficit Thresholds

Edit the deficit_thresholds array in CONFIG to test different water stress levels:

deficit_thresholds: [-10, -30, -50, -80, -120, -200],  // Custom thresholds

Lower (more negative) values = stricter criterion = fewer suitable dekads.

Analyzing Different Years

Change the year parameter:

year: 2020,  // Or 2019, 2021, etc.

The script will automatically adjust all date filters.

Adjusting Spatial Resolution

Change output_scale for coarser/finer output:

output_scale: 1000,  // 1 km resolution (faster)
// or
output_scale: 250,   // 250 m resolution (slower, more detail)

Higher resolution increases computation time exponentially.

Using Different Rice Classification

Replace the rice_map definition to use alternative data sources:

// Option 1: ESA WorldCover 2021 (global, high quality)
var worldcover = ee.ImageCollection("ESA/WorldCover/v200")
  .filterDate("2021-01-01", "2021-12-31")
  .first()
  .select("Map");
var rice_map = worldcover.eq(40);  // Class 40 = Croplands

// Option 2: MODIS LC (annual, consistent time series)
var modis_lc = ee.ImageCollection("MODIS/061/MCD12Q1")
  .filterDate(CONFIG.year + "-01-01", CONFIG.year + "-12-31")
  .first()
  .select("LC_Type1");
var rice_map = modis_lc.eq(10);  // Class 10 = Croplands

// Option 3: Custom asset (best for regional studies)
var rice_map = ee.Image("users/your-username/japan_rice_paddy_map_2020");

Troubleshooting

Issue: "Image is null" or "No valid data"

Causes:

Study area boundary doesn't overlap with rice extent
Year has no data for selected collections
Bounding box coordinates are inverted

Solutions:

Verify bounding box: [min_lon, min_lat, max_lon, max_lat]
Check that min_lon < max_lon and min_lat < max_lat
Try expanding the study area boundary
Verify year is within CHIRPS coverage (1981-present)

Issue: "Export failed" or "Computation timed out"

Causes:

Study area too large (>1e13 pixels at 500m = ~2.5 million km²)
Output resolution too fine (250m with large area)
Too many thresholds being processed

Solutions:

Reduce study area or increase output_scale (coarsen resolution)
Reduce number of thresholds in deficit_thresholds
Split analysis into smaller regions
Use export.image.toDrive with default pyramid policy

Issue: "Authentication failed" or "Access denied"

Causes:

Not logged into Google Earth Engine
Missing access to shared datasets

Solutions:

Click "Authorize" when prompted by GEE
Ensure you have a Google Earth Engine account (request access at https://earthengine.google.com/signup/)
If using user assets, verify ownership

Performance Optimization

Reduce Computation Time

// Option 1: Reduce spatial resolution
output_scale: 1000  // Default 500m, use 1000m for faster preview

// Option 2: Reduce study area
var study_region = ee.Geometry.rectangle([130, 30, 135, 35]);  // Smaller area

// Option 3: Use fewer thresholds
deficit_thresholds: [-50, -100, -150]  // Instead of 7 thresholds

Monitor Progress

Look at the GEE console (bottom-right) to see:

Number of tiles being processed
Current dekad being analyzed
Estimated remaining time

Integration with Python Pipeline

After downloading GeoTIFF from Google Drive:

Place in data/raw/gee_exports/
Run Python pipeline:
```
python scripts/run_pipeline.py
```
Pipeline will automatically load and post-process the GEE outputs

Support & Documentation

GEE Documentation: https://developers.google.com/earth-engine
GEE Datasets: https://developers.google.com/earth-engine/datasets
CHIRPS Data Guide: https://www.chg.ucsb.edu/data/chirps
MODIS PET Guide: https://lpdaac.usgs.gov/products/mod16a2v061/
SoilGrids Info: https://www.isric.org/explore/soilgrids

Citation

If you use this GEE script, please cite:

@software{tran_awd_gee_2025,
  title={Google Earth Engine Pipeline for AWD Policy Transportability Assessment},
  author={Tran, My},
  year={2025},
  url={https://github.com/MyTran-GitHub/awd-policy-transportability/tree/main/gee}
}

Last Updated: January 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Google Earth Engine Setup Guide

Overview

Prerequisites

Quick Start

1. Open Script in GEE Editor

2. Configure Study Area

3. Load Study Area Data

4. Run the Script

5. Download Results

Output Specification

File Format

Interpretation

Data Sources

1. Rainfall: CHIRPS

2. Evapotranspiration: MODIS MOD16A2

3. Soil Properties: SoilGrids

4. Elevation: SRTM

5. Rice Extent Map

Advanced Configuration

Changing Deficit Thresholds

Analyzing Different Years

Adjusting Spatial Resolution

Using Different Rice Classification

Troubleshooting

Issue: "Image is null" or "No valid data"

Issue: "Export failed" or "Computation timed out"

Issue: "Authentication failed" or "Access denied"

Performance Optimization

Reduce Computation Time

Monitor Progress

Integration with Python Pipeline

Support & Documentation

Citation

FilesExpand file tree

README.md

Latest commit

History

README.md

File metadata and controls

Google Earth Engine Setup Guide

Overview

Prerequisites

Quick Start

1. Open Script in GEE Editor

2. Configure Study Area

3. Load Study Area Data

4. Run the Script

5. Download Results

Output Specification

File Format

Interpretation

Data Sources

1. Rainfall: CHIRPS

2. Evapotranspiration: MODIS MOD16A2

3. Soil Properties: SoilGrids

4. Elevation: SRTM

5. Rice Extent Map

Advanced Configuration

Changing Deficit Thresholds

Analyzing Different Years

Adjusting Spatial Resolution

Using Different Rice Classification

Troubleshooting

Issue: "Image is null" or "No valid data"

Issue: "Export failed" or "Computation timed out"

Issue: "Authentication failed" or "Access denied"

Performance Optimization

Reduce Computation Time

Monitor Progress

Integration with Python Pipeline

Support & Documentation

Citation